Application Identification using Supervised Clustering Method

The classification of traffic provides essential data for network management and research. Several classification approaches are developed and proposed to protect the network resources or enforce organizational policies. Whereas the port number based classification works only for some well-known applications and payload based classification is not suitable for encrypted packet payloads that make the sense to classify the traffic based on behaviors observed in networks. In this paper, a supervised clustering algorithm called Flow Level based Classification (FLC) is proposed to classify network flows, which comprises of flows in the same conversation. In this paper we discussed recent laurels and various research trends in supervised and unsupervised clustering algorithms. We outline the obstinately mysterious challenges in the field over the last decade and suggest strategies for tackling these challenges to promote headway in the art of Internet traffic classification.